Use numpy bincount with weights for 2d array-CodePudding

Suppose I have the following 1d numpy arrays

>>> x = np.array([1,2,3,4,5,6])
>>> y = np.array([10,20,30,40,50,60])

and bins

>>> bins = np.array([0,0,1,1,2,3])

Then I may use bincount for each array like:

>>> np.bincount(bins, weights=x)
array([ 3.,  7.,  5.,  6.])
>>> np.bincount(bins, weights=y)
array([ 30.,  70.,  50.,  60.])

Can I perform both summaries at once ? I tried

>>> np.bincount(np.array([[0,0,1,1,2,3], [0,0,1,1,2,3]]), weights=np.array([[1,2,3,4,5,6], [10,20,30,40,50,60]]))
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
ValueError: object too deep for desired array
>>>

and got an error

CodePudding user response：

Nope, it's not possible. And one hint is that there isn't any axis argument that you can pass to bincounts. Your best bet is do them one-at-a-time:

bc = np.array([np.bincount(bins, weights=x), np.bincount(bins, weights=y)])

Or using list-comprehension:

bc = np.array([np.bincount(bins, weights=w) for w in [x, y]])

Output:

>>> bc
array([[ 3.,  7.,  5.,  6.],
       [30., 70., 50., 60.]])

CodePudding user response：

You can simulate bincount with np.add.at:

xy = np.stack([x,y])
n = np.unique(bins).size
out = np.zeros((2,n), dtype=xy.dtype)
np.add.at(out, (slice(None), bins), xy)

output in out:

array([[ 3,  7,  5,  6],
       [30, 70, 50, 60]])